Incentivos na atenção primária de saúde brasileira e seus efeitos sobre mortalidade e morbidade

Abstract

The objective of this study is to evaluate the impact of the performance compensation program of the brazilian health system (Sistema Único de Saúde) - the PMAQ - on municipal indicators of mortality and morbidity that are sensitive to variations in the primary health care supply. The sample used is restricted to the municipalities participating in the Family Health Strategy during the period from 2010 to 2015, in order to capture the complementary effect of the PMAQ. For this, regressions are estimated using the fixed effects estimator combined with the propensity score matching method. The results indicate that the program had mixed effects on the indicators of mortality and morbidity of the participating municipalities. The estimates show positive effects on reducing infant and child mortality rates due to endocrine, nutritional and metabolic diseases, as well as reducing infant mortality rates due to infectious and parasitic diseases and respiratory diseases. However, some adverse effects are found on the mortality rates of children and adults for some causes. The results also indicate that the program had no significant effect on the mortality rates of the elderly. Regarding morbidity, the estimates suggest that the program did not affect the reduction of most of the selected conditions, and that some adverse effects are also found. There is evidence that PMAQ has a positive impact on the reduction of general hospitalizations due to pulmonary diseases, infantile hospitalizations due to epilepsy, hospitalizations of children due to anemia and nutritional deficiencies, and hospitalizations of adults due to immunizable and pulmonary diseases. The estimates show more significant results for the age group of the elderly, indicating that PMAQ positively impacted the reduction of hospitalization rates due to gastroenteritis, pulmonary diseases, epilepsy and skin and subcutaneous tissue diseases. In addition, unmatched regressions are estimated and placebo tests are performed for comparison and robustness checks.